Reduction of False Positives in the Identification of Nuclides in Environmental Samples

Gamma-ray spectra containing peaks that are too close in energy for deconvolution to be done accurately are best analyzed by a library-based method. However, a library-based analysis done with a large library may result an unacceptable number of false positives being reported. A normal working library suitable for analyzing environmental samples containing unknown materials may have over 1000 peaks, many of which are too close for deconvolution to be done accurately. A program has been written that uses a library-based analysis method that reduces the reporting of false positives, while retaining the ability to identify isotopes accurately from a large range of possibilities. In addition, the peak area calculation has been improved by allowing the energy calibration to be a free parameter in the fit of individual multiplets. This peak area improvement can result in a change of activity of several percent for some nuclides. In some cases, shifting the multiplet position can reduce false positives by identifying a peak in the multiplet as an unknown rather than associating it with a nuclide. One spectrum showed a marked peak activity improvement when the calibration was allowed to shift even though the the average error in the original calibration was 0.08%. Results obtained from a study on the identification of uranium decay products are presented.